[ace-ify]

What this changes

Shows tool call details (inputs and outputs) in collaboration channels safely with secret redaction, size bounding, and room-level disclosure controls:

• Secret Redaction (switch_core.bridges.collaboration.session.redaction): Masks sensitive credentials including PEM private keys, provider API keys (OpenAI, Anthropic, GitHub, Slack, AWS, Google), JWTs, Bearer tokens, database connection string passwords, MCP/HTTP Authorization headers, and environment variable secret assignments before channel rendering.
• Bounded Detail Previews: Enforces limits (up to 500 chars for input, 2,000 chars total) with descriptive omission notices so verbose tool outputs don't flood chat channels.
• Room-Level Privacy Control (RoomStore): Supports tool_detail_off room metadata toggle (tool_detail_enabled_in, tool_detail_enabled, set_tool_detail_enabled). Stored execution detail is preserved in DB and stripped on read/render when the toggle is off.
• Renderer Integration: Plumbed safe redaction and toggle support into Slack (_plan_task, _plan) and neutral/Mattermost (activity_log, _log_line) collaboration session renderers.

Closes #803

How it was tested

1. Added unit test suite tests/switch_core/bridges/collaboration/test_safe_tool_redaction.py (10 tests) covering PEM keys, provider tokens, DB URLs, headers, env secrets, and character limit truncation.
2. Added rendering test suite tests/switch_core/bridges/collaboration/test_safe_tool_rendering.py (5 tests) verifying Slack card detail redaction, neutral activity log masking, and room toggle state behavior.
3. Verified against existing collaboration bridge suites (test_session_slack_requests.py).
4. Verified lint and formatting compliance with ruff check and ruff format.

============================= 49 passed in 0.43s ==============================

[github-actions]

All contributors have signed the CLA. ✅
_{Posted by the CLA Assistant Lite bot.}

[ace-ify]

I have read the CLA Document and I hereby sign the CLA

[ace-ify]

recheck